仓库/Skill 库

[ICML 2026] 一种神经符号框架，用于溯因推理，将多智能体协作锚定在基于因果图与状态机的结构化信念状态上。[ICML 2026] A neuro-symbolic framework for abductive reasoning that grounds multi-agent collaboration in structured belief states using a causal graph and state machine.

[EMNLP 2026] LightRSI：一个模块化运行时，用于在长程 LLM agents 中轻量部署递归自改进循环。[EMNLP 2026] LightRSI is a modular runtime for lightly deploying recursive self-improvement loops in long-horizon LLM agents.

Promptfoo 的 GitHub Action。可用于测试 prompt、Agent 与 RAG；支持 LLM 的 AI 红队对抗、渗透测试与漏洞扫描；对比 GPT、Claude、Gemini、Llama 等模型的性能；通过简洁的声明式配置集成命令行与 CI/CD。The GitHub Action for Promptfoo. Test your prompts, agents, and RAGs. AI Red teaming, pentesting, and vulnerability scanning for LLMs. Compare performance of GPT, Claude, Gemini, Llama, and more. Simple declarative configs with command line and CI/CD integration.

Whip 是一个面向 Agent 的原生 SSH 移动客户端，从一开始便专为 Herdr 多路复用器设计。Whip is an agent-native SSH mobile client, designed specifically for the Herdr multiplexer from day one

AI-ML Companion：一个通过"看 AI 运作"来学习 AI 与 ML 的交互式平台 — 28 条学习路径、400+ 模块、实时可视化 — 另有 20+ 端到端真实项目，覆盖经典 ML、深度学习、计算机视觉到生产级 LLM 与多 Agent 系统。AI-ML Companion: an interactive platform to learn AI & ML by watching it work - 28 tracks, 400+ modules, live visualizations - plus 20+ end-to-end real world projects from classical ML, DL, Computer Vision to production LLM & multi-agent systems.

动手学大模型全栈：CS336 中文精讲 · PyTorch 手搓 Transformer · 单卡复现 Pretrain/SFT/LoRA/DPO/GRPO/RLVR · PEFT/Agent/RAG 落地（含真实实验、曲线与复现脚本）

开源与闭源 AI agent 编排——一套完全可定制的 AI agent 队列，覆盖从头脑风暴到完全自主的项目维护各个层级AI agent orchestration for open and closed source — a fully customizable fleet of AI agents covering every level of project maintenance, from brainstorming to full autonomy

Intent as Code｜AI 的工作流语言。一个文件、4 个动词、一个 Rust 二进制程序。本地优先，支持任意模型，AGPL-3.0。🦋Intent as Code | the workflow language for AI. One file, 4 verbs, one Rust binary. Local-first, any model, AGPL-3.0. 🦋

真实 AI Agent 安全事件、入侵和漏洞的精选时间线（2024-2026）。每条记录均标注来源和日期。A curated timeline of real AI agent security incidents, breaches, and vulnerabilities (2024-2026). Every entry sourced and dated.

Vibe first. Then FastAgent——将本地 Agent 目录转化为实时服务：嵌入你的应用、部署到 GitHub、接入 Telegram 或任意渠道。Vibe first. Then FastAgent — turn a local agent directory into a live service: in your app, on GitHub, in Telegram, or any channel.

在你自有的服务器上运行一家由 Agent 组成"公司"。可启动具名 Agent（claude、codex、pi 等），将它们组织成组织架构图并共享待办列表，让它们交接工作，只在需要人类决策时推送通知到你的手机。MIT 协议。Run a company of AI agents on a server you own. Spin up named agents (claude, codex, pi…), put them on an org chart with a shared backlog, let them hand off work and ping your phone only when a human must decide. MIT.

通过 hooks 拦截 AI 编程 Agent 发起的 cargo/gcc 构建，并透明地将其路由到远程工作机器，像在本地编译一样返回产物。Intercepts cargo/gcc builds from AI coding agents via hooks and transparently routes them to remote worker machines, returning artifacts as if compiled locally

面向 OpenCode 与 Claude 的轻量级持久记忆伴侣A lightweight persistent-memory companion for OpenCode & Claude

为 AI 编码 Agent 提供长期、跨项目的记忆。以你自己的 Obsidian 仓库为唯一信源。无守护进程、不依赖不透明数据库，记忆归你所有。Long-term, cross-project memory for AI coding agents. Your own Obsidian vault as the source of truth. Daemonless and without opaque databases, your memory belongs to you.

开源 AI CLI 与本地 MCP 服务器，连接 25 个 AI 客户端（包括 Claude Code、Cursor、Codex 和 ChatGPT）与 2000+ 模型和 API，支持 OAuth 与回滚Open-source AI CLI and local MCP server connecting 25 AI clients—including Claude Code, Cursor, Codex, and ChatGPT—to 2,000+ models and APIs, with OAuth and rollback.

即插即用、成本高效且轻量化的长周期 LLM Agent 记忆插件A Plug-and-Play, Cost-Efficient, and Lightweight Memory Plugin for Long-Horizon LLM Agents

AI Agent 就绪的 CLI 框架，用于 Excel VBA 项目的编辑、测试、运行、追踪与差异比对。AI-Agent-ready CLI framework for editing, testing, running, tracing, and diffing Excel VBA projects.

AI 编程智能体统一桌面工作台 | Desktop hub for managing Claude Code, Codex CLI, Gemini CLI & OpenCode. Tauri 2 · Rust · React · SQLite.

面向编程 Agent 的更快运行时。平均提速 25%、降本 30%，同时质量持平或更优。同等任务、同等质量，更快更便宜。Faster runtime for coding agents. Make coding agents 25% faster and 30% cheaper on average while keeping the quality same or more. Same Task, Same Quality, Faster and Cheaper.

无头持久化 Agent 调度引擎，可可靠且可扩展地运行持久化 AI Agent。Headless durable agentic harness engine. Run durable AI agents reliably and scalably.

用于编排 AI 编程 Agent 的委派环境。支持跨 git worktree 管理 Claude、Gemini 与 Codex 会话，集成终端、上下文注入与工作流自动化。A delegation environment for orchestrating AI coding agents. Manage Claude, Gemini, and Codex sessions across git worktrees with integrated terminals, context injection, and workflow automation.

开源、自托管的 Linear 替代方案。项目管理与 issue tracking，团队与 AI Agent 并肩协作完成产品规划与交付。Open-source, self-hosted alternative to Linear. Project management and issue tracking where teams and AI agents work side by side to plan and ship products.

Zorv AI — 安卓 Android 端开源 AI Agent 智能体助手。多模型对话、人格系统、记忆、语音全双工 TTS/STT、定时任务、可扩展工具链、飞书/QQ/微信接入。Kotlin + Jetpack Compose 开发。

开源元搜索引擎后端、MCP server 与面向 LLM agent 的 AI 搜索 API。基于 Python FastAPI 的搜索网关，通过 SerpBase 与 Serper 集成 Google 搜索，支持多引擎搜索聚合、结构化 JSON 输出、provider 回退与去重，并提供 SearXNG 替代架构以适配 agent 工作流。Open-source metasearch backend, MCP server, and AI search API for LLM agents. Python FastAPI search gateway with Google search via SerpBase and Serper, multi-engine search aggregation, structured JSON output, provider fallback, deduplication, and SearXNG alternative architecture for agent workflows.

面向 WordPress 的 SEO 内容创作与 Humanizer 工具，基于 OpenAI、Anthropic Claude 和 Gemini，支持可编辑 prompt、图片与 Opace AI Hub。SEO content creator and Humanizer for WordPress using OpenAI, Anthropic Claude and Gemini, with editable prompts, images and Opace AI Hub.

Agentic 预测市场交易的 Python SDKPython SDK for Agentic Prediction Market trading

Claude Code workflow skills——GitHub issue/PR/release 自动化与基于 Fable 的规划。可通过 npx 或作为 Claude Code 插件安装。Claude Code workflow skills — GitHub issue/PR/release automation and Fable-driven planning. Install via npx or as a Claude Code plugin.

AI 编写软件的记录系统。实体、关系、变更与来源的持久化图谱，使人类和 AI Agent 在合并前看到变更影响范围。目前与 Git 并行使用。The system of record for AI-written software. A persistent graph of entities, relationships, changes, and provenance, so humans and AI agents see what a change touches before it merges. Beside Git today.

基于 cocoon 的快速冷启动 MicroVM 沙箱，为 AI Agent 提供隔离运行环境。Fast cold-boot MicroVM sandboxes for AI agents on cocoon

生产级多 Agent 编排引擎。从可插拔的 Agent、LLM、工具与 retriever 目录组合 Agentic 工作流；本地使用 LangGraph 执行，或通过 Temporal 分布式执行。内置 RAG 管道用于企业知识检索，支持 A2A 与 MCP 协议，提供可视化拖拽蓝图构建器。Production-grade multi-agent orchestration engine. Compose agentic workflows from a pluggable catalog of Agents, LLMs, tools, and retrievers. Execute locally with LangGraph or distributed with Temporal. Built-in RAG pipeline for enterprise knowledge retrieval. A2A and MCP protocol support. Visual drag-and-drop blueprint builder.

自托管、自进化的联邦化全能应用——通过 Tailscale 在单一仪表板上提供开发管理、Agent 编排、持久心智、数字孪生、第二大脑以及创意流水线，将用户接入操作系统。Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS

开源 AI Workforce OS：将 AI Agent 转化为执行业务职能、以结果衡量的自主劳动力。Claude Tag 的开源替代方案。The Open-Source AI Workforce OS. Turn AI agents into autonomous labor that performs business functions and is measured by outcomes. Claude Tag open source alternative

智能多 agent 对话助手，采用 LangGraph 编排，支持 Human-in-the-Loop、企业级可观测性与完整 i18n 支持（6 种语言）。Smart multi-agent conversational assistant with LangGraph orchestration, Human-in-the-Loop, enterprise-grade observability, and full i18n support (6 languages)

工程化组织结构，而非仅仅工程化 Agent。601 个精选资源 · 9 个设计层级 · 11 个章节 · 279 篇论文与预印本 —— 一份面向图结构多 Agent 系统（角色、拓扑、交接、工作图、状态、闸门、可靠性、可观测性）的领域指南、CC0 开源数据集与交互式图谱。🕸️ Engineer the organization, not just the agent. 601 curated resources · 9 design layers · 11 sections · 279 papers & preprints — a field guide, CC0 open dataset, and interactive atlas for graph-structured multi-agent systems: roles, topologies, handoffs, work graphs, state, gates, reliability, observability.

一种基于大语言模型、采用人在回路的科学模式挖掘工作流A Human-in-the-Loop Workflow for Scientific Schema Mining with Large Language Models

📚 AI 驱动的论文阅读工作台。基于项目的论文问答，集成 Hybrid RAG、多 Agent 工作流（ReAct/Plan-Act/RePlan）、长期记忆与可溯源证据。使用 LangChain + LangGraph + Streamlit 构建。📚 AI-powered research reading workbench. Project-based paper Q&A with Hybrid RAG, multi-agent workflows (ReAct/Plan-Act/RePlan), long-term memory, and traceable evidence. Built with LangChain + LangGraph + Streamlit.